Benefits and Harms of Oral Anticoagulant Therapy in Chronic Kidney Disease

Chronic kidney disease (CKD) is a prothrombotic state that is associated with substantially increased risks for arterial and venous thromboembolism (VTE) (1). In addition, atrial fibrillation (AF) is highly prevalent in this population, affecting 18% of patients with CKD (2) and 12% to 25% of those with dialysis-dependent end-stage kidney disease (ESKD) (3, 4). The presence of CKD increases risks for stroke or systemic embolism, congestive heart failure, myocardial infarction, and all-cause death among patients with AF (5, 6). Compared with persons with normal kidney function, risk for VTE is almost 2-fold greater among those with an estimated glomerular filtration rate (eGFR) between 15 and 59 mL/min/1.73 m2 (7) and 3-fold greater in those with dialysis-dependent ESKD (8). Venous thromboembolism in ESKD is also associated with increased risks for bleeding and all-cause death (8). Other common clinical manifestations of increased thrombotic risk in CKD include acute coronary syndrome, stroke, peripheral artery occlusion, and dialysis access thrombosis (1, 9). Anticoagulant therapy is an important intervention in the prevention of cardiovascular thrombotic and VTE events. Evidence-based treatment guidelines recommend anticoagulation for prevention of stroke in patients with nonvalvular AF and a CHA2DS2-VASc score of 2 or greater in men or 3 or greater in women (10, 11), for VTE in patients who have had major orthopedic or nonorthopedic surgery or hospitalized patients with acute illness (12), and for recurrent VTE in patients with VTE disease (13). Patients with advanced CKD and ESKD who have AF are prescribed oral anticoagulant (OAC) therapy less frequently than those with normal kidney function (3, 14). Use of warfarin in patients receiving dialysis who have AF varies from 2% in Germany to 37% in Canada (3). The low rates of anticoagulant therapy use in advanced CKD and ESKD may be due to the increased risk for bleeding, uncertainty about potential benefits in this population, warfarin-associated calciphylaxis, and warfarin-related nephropathy (15, 16). In CKD, risk for major bleeding increases linearly with decreasing eGFR (17). In patients with dialysis-dependent ESKD, bleeding risk is further increased with incremental use of antithrombotic agents, such as warfarin and antiplatelets (18). The exclusion of patients with CKD from nearly 90% of trials evaluating anticoagulants has contributed to uncertainty about the role of anticoagulant therapy in CKD (19). The aim of the current systematic review was to evaluate the benefits and harms of OAC therapy for a range of clinical indications in patients with CKD stages 3 to 5, including those receiving dialysis. Methods This systematic review and meta-analysis was conducted according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) statement (20). The protocol for this review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO) on 4 December 2017 (www.crd.york.ac.uk/prospero/display_record.php?RecordID=79709). Data Sources and Searches Relevant studies were identified by performing English-language searches of MEDLINE (inception to February 2019), EMBASE (inception to February 2019), and the Cochrane Central Register of Controlled Trials (January 2019) using the search strategy described in Supplement Table 1. In addition, reference lists of relevant systematic reviews were searched. ClinicalTrials.gov was searched (25 February 2019) using the following terms: chronic kidney disease, renal dialysis, atrial fibrillation, and anticoagulation. Supplement. Supplemental Material Study Selection and Outcomes Studies were eligible for inclusion if they were randomized controlled trials; included adults with CKD (creatinine clearance [CrCl] <60 mL/min or eGFR <60 mL/min/1.73 m2) or dialysis-dependent ESKD; compared a vitamin K antagonist (VKA) or nonvitamin K oral anticoagulant (NOAC) with another OAC, placebo, low-molecular-weight heparin (LMWH), aspirin, or no study medication; and reported efficacy, bleeding outcomes, or both. All indications for anticoagulation were eligible for inclusion. Two authors (J.T.H. and B.L.N.) independently reviewed each title and abstract and reviewed the full texts of shortlisted studies. Disagreements about study eligibility were resolved via consultation with 2 other authors (V.P. and S.V.B.). If multiple secondary publications of the same trial were identified, the one with the most complete data was used and additional data from secondary sources were extracted. Incomplete or unpublished trial data were requested from the investigators. The outcomes of this systematic review were stroke or systemic embolism in AF, nonhemorrhagic stroke, hemorrhagic stroke, all-cause or cardiovascular death, VTE or VTE-related death, myocardial infarction, composite cardiovascular events (cardiovascular or all-cause death, nonfatal myocardial infarction, or stroke), dialysis access thrombotic events, major bleeding, major or nonmajor clinically relevant bleeding, and intracranial hemorrhage. Data Extraction and Quality Assessment Data were extracted independently by 2 authors (J.T.H. and B.L.N.), and disagreements were resolved via consultation with 2 other authors (V.P. and S.V.B.). A standardized form was used to extract the following data: patient demographic characteristics, study design and conduct, indication for anticoagulation, drug dose, nonrandomized co-interventions, follow-up duration, and outcome and bleeding events. The methodological quality of each included study was assessed at the outcome level independently by 2 authors (J.T.H. and B.L.N.) using the risk-of-bias assessment tool developed by the Cochrane Bias Methods Group (21). Data Synthesis and Analysis The results were expressed as risk ratios (RRs) with 95% CIs. A treatment group continuity correction was used if there were 0 events in 1 group in a trial. For trials with 3 groups comparing 2 different doses of NOACs with VKAs, data from only the high-dose NOAC groups were used for the main analyses to avoid potentially uninterpretable results caused by merging of the benefits and harms of different doses; this was similar to the method used in a previous meta-analysis (22). Additional analyses were conducted by combining data from both high- and low-dose groups of NOACs. Summary estimates were obtained with a random-effects model using the PauleMandel method (23). If data on the number of events and participants were not reported, a generic inverse variance meta-analysis was performed by calculating the log of the hazard ratio and its SE from the reported hazard ratio and its CI. Statistical heterogeneity across studies was estimated using the I 2 test, with values of 25%, 50%, and 75% corresponding to low, moderate, and high heterogeneity, respectively (24). Statistical analyses were performed using Stata/MP, version 15.1 (StataCorp), and R, version 3.5.3 (R Foundation for Statistical Computing). Using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach, 3 authors (J.T.H., B.L.N., and L.P.C.) summarized the certainty of the evidence based on the following domains: within-study risk of bias, indirectness of evidence, unexplained heterogeneity or inconsistency of results, and imprecision of results. Disagreements were resolved via consultation with 2 other authors (M.J. and S.V.B.) (25). Because all meta-analyses involved fewer than 10 trials, small-study effects (publication bias) were not assessed and publication bias was not included in ratings of certainty of evidence (26). Role of the Funding Source This study received no funding. Results Selection and Description of Studies Forty-five trials that involved 34082 participants and evaluated VKAs or NOACs were included in the systematic review (median sample size, 276 participants [range, 10 to 4168 participants]; median follow-up, 12 months [range, 1 to 36 months]) (Figure 1). Of these trials, 8 included 685 participants with dialysis-dependent ESKD (median sample size, 91 participants [range, 18 to 174 participants]; median follow-up, 12 months [range, 3 to 36 months]), with 7 evaluating VKAs for prevention of dialysis access thrombosis and 1 evaluating the effect of VKAs on hemostatic factors. The remaining 37 trials included 33397 participants with CKD who were not receiving dialysis (defined as CrCl of 20 to 60 mL/min, eGFR of 15 to 60 mL/min/1.73 m2, or serum creatinine level 1.5 mg/dL; median sample size, 380 participants [range, 10 to 4168 participants]; median follow-up, 12 months [range, 1 to 36 months]). Eleven trials included 16787 participants with AF (median sample size, 516 participants [range, 12 to 4074 participants]; median follow-up, 14 months [range, 3 to 34 months]). Eleven trials involved 2975 participants with acute VTE (median sample size, 162 participants [range, 72 to 657 participants]; median follow-up, 12 months [range, 6 to 36 months]). Six trials included 3908 medically ill or perioperative participants requiring anticoagulation for thromboprophylaxis (median sample size, 380 participants [range, 42 to 2197 participants]; median follow-up, 2 months [range, 1 to 6 months]). The remaining 9 trials involved 9727 participants with cardiovascular disease other than AF (median sample size, 331 participants [range, 72 to 4168 participants]; median follow-up, 9 months [range, 1 to 36 months]). Data from the 37 trials involving patients with nondialysis CKD were obtained exclusively from CKD subgroup analyses of large trials. Details of the included trials are provided in Supplement Table 2. Figure 1. Evidence search and selection. AF = atrial fibrillation; CAD = coronary artery disease; ESKD = end-stage kidney disease; PAD = peripheral artery disease; RCT = randomized controlled trial; VTE = venous thromboembolism. Nonvitamin K oral anticoagulants were compared with VKAs (15 trials, 1649